Friends or Neighbors? Advice Networks, Geographic Proximity, and the Transmission of Health Knowledge

Public-health interventions often aspire to generate spillovers beyond the households they directly reach, but identifying who is exposed to whom within communities is challenging. Researchers frequently approximate social exposure using geographic proximity, even though the individuals who transmit health advice need not be nearby. We study this issue by conducting secondary analysis of a randomized maternal and child health counseling intervention in rural Honduras with geocoded households and mapped social networks. At our main 50-meter radius, geographic proximity captures only 30.8% of health-advice ties, while only 3.7% of geographically proximate household pairs are connected by advice-giving. The choice of exposure mapping matters for what the analysis detects. Exposure mappings based solely on geographic proximity yield a precise zero effect on diarrhea knowledge. By contrast, exposure through advice ties beyond 50 meters increases diarrhea knowledge, and the effect is statistically distinguishable from exposure through nearby non-advice contacts. The evidence suggests that simple, actionable illness-management knowledge diffuses through advice relationships that geographic proximity alone fails to capture. Reported diarrhea, indicating whether the respondent or any child had diarrhea in the previous four weeks, also increases with advice exposure outside 50m, consistent with greater recognition, recall, or reporting rather than higher underlying morbidity. We find no corresponding effects for respiratory knowledge or reported respiratory illness. Our results show that using geographic proximity as a proxy for social exposure can mask important channels of information transmission and could lead researchers to miss meaningful spillover effects.